Convolutional Neural Networks

Project: Write an Algorithm for a Dog Identification App


In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the Jupyter Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.

In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this Jupyter notebook.


Why We're Here

In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).

Sample Dog Output

In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!

The Road Ahead

We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.

  • Step 0: Import Datasets
  • Step 1: Detect Humans
  • Step 2: Detect Dogs
  • Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
  • Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)
  • Step 5: Write your Algorithm
  • Step 6: Test Your Algorithm

Step 0: Import Datasets

Make sure that you've downloaded the required human and dog datasets:

  • Download the dog dataset. Unzip the folder and place it in this project's home directory, at the location /dogImages.

  • Download the human dataset. Unzip the folder and place it in the home directory, at location /lfw.

Note: If you are using a Windows machine, you are encouraged to use 7zip to extract the folder.

In the code cell below, we save the file paths for both the human (LFW) dataset and dog dataset in the numpy arrays human_files and dog_files.

In [1]:
import numpy as np
from glob import glob

# load filenames for human and dog images
human_files = np.array(glob("lfw/*/*"))
dog_files = np.array(glob("dogImages/*/*/*"))

# print number of images in each dataset
print('There are %d total human images.' % len(human_files))
print('There are %d total dog images.' % len(dog_files))
There are 13233 total human images.
There are 8351 total dog images.

Step 1: Detect Humans

In this section, we use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images.

OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory. In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.

In [2]:
import cv2                
import matplotlib.pyplot as plt                        
%matplotlib inline                               

# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')

# load color (BGR) image
img = cv2.imread(human_files[0])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# find faces in image
faces = face_cascade.detectMultiScale(gray)

# print number of faces detected in the image
print('Number of faces detected:', len(faces))

# get bounding box for each detected face
for (x,y,w,h) in faces:
    # add bounding box to color image
    cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
    
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()
Number of faces detected: 1

Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.

Write a Human Face Detector

We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.

In [3]:
# returns "True" if face is detected in image stored at img_path
def face_detector(img_path):
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray)
    return len(faces) > 0

(IMPLEMENTATION) Assess the Human Face Detector

Question 1: Use the code cell below to test the performance of the face_detector function.

  • What percentage of the first 100 images in human_files have a detected human face?
  • What percentage of the first 100 images in dog_files have a detected human face?

Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.

Answer: (You can print out your results and/or write your percentages in this cell)

In [4]:
from tqdm import tqdm

human_files_short = human_files[:100]
dog_files_short = dog_files[:100]

#-#-# Do NOT modify the code above this line. #-#-#

## TODO: Test the performance of the face_detector algorithm 
## on the images in human_files_short and dog_files_short.

def test_detector(files):
    count = 0
    total = float(files.size) 
    for filename in files:
        if face_detector(filename):
            count = count + 1
    return(count/total)

print(str(test_detector(human_files_short)*100)+"% of human_files_short detected as human face")
print(str(test_detector(dog_files_short)*100) + "% of dog_files_short detected as human face")
100.0% of human_files_short detected as human face
9.0% of dog_files_short detected as human face

We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on human_files_short and dog_files_short.

In [5]:
### (Optional) 
### TODO: Test performance of another face detection algorithm.
### Feel free to use as many code cells as needed.

Step 2: Detect Dogs

In this section, we use a pre-trained model to detect dogs in images.

Obtain Pre-trained VGG-16 Model

The code cell below downloads the VGG-16 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories.

In [6]:
import torch
import torchvision.models as models

# define VGG16 model
VGG16 = models.vgg16(pretrained=True)

# check if CUDA is available
use_cuda = torch.cuda.is_available()

# move model to GPU if CUDA is available
if use_cuda:
    VGG16 = VGG16.cuda()
Downloading: "https://download.pytorch.org/models/vgg16-397923af.pth" to /home/ubuntu/.cache/torch/checkpoints/vgg16-397923af.pth
100%|██████████| 528M/528M [00:18<00:00, 29.8MB/s] 

Given an image, this pre-trained VGG-16 model returns a prediction (derived from the 1000 possible categories in ImageNet) for the object that is contained in the image.

(IMPLEMENTATION) Making Predictions with a Pre-trained Model

In the next code cell, you will write a function that accepts a path to an image (such as 'dogImages/train/001.Affenpinscher/Affenpinscher_00001.jpg') as input and returns the index corresponding to the ImageNet class that is predicted by the pre-trained VGG-16 model. The output should always be an integer between 0 and 999, inclusive.

Before writing the function, make sure that you take the time to learn how to appropriately pre-process tensors for pre-trained models in the PyTorch documentation.

In [7]:
from PIL import Image
import torchvision.transforms as transforms

# Set PIL to be tolerant of image files that are truncated.
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

def VGG16_predict(img_path):
    '''
    Use pre-trained VGG-16 model to obtain index corresponding to 
    predicted ImageNet class for image at specified path
    
    Args:
        img_path: path to an image
        
    Returns:
        Index corresponding to VGG-16 model's prediction
    '''
    
    ## TODO: Complete the function.
    ## Load and pre-process an image from the given img_path
    ## Return the *index* of the predicted class for that image
    
    
    transform = transforms.Compose([
        transforms.Resize(256),               
        transforms.CenterCrop(224),           
        transforms.ToTensor(),                
        transforms.Normalize(                 
            mean=[0.485, 0.456, 0.406],           
            std=[0.229, 0.224, 0.225]             
        )])
    
    image = Image.open(img_path)    
    image = transform(image)
    
    if use_cuda:
        image = image.cuda()
    
    batch = torch.unsqueeze(image, 0)
        
    output = VGG16(batch) 
    
    _, index = torch.max(output, 1)
        
    return int(index) # predicted class index

(IMPLEMENTATION) Write a Dog Detector

While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained VGG-16 model, we need only check if the pre-trained model predicts an index between 151 and 268 (inclusive).

Use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).

In [8]:
### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
    ## TODO: Complete the function.    
    
    index = VGG16_predict(img_path)
    
    if index >= 151 and index <=268:
        return True
    
    return False

(IMPLEMENTATION) Assess the Dog Detector

Question 2: Use the code cell below to test the performance of your dog_detector function.

  • What percentage of the images in human_files_short have a detected dog?
  • What percentage of the images in dog_files_short have a detected dog?

Answer:

In [9]:
### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.

def test_dog_detector(files):
    count = 0
    total = float(files.size) 
    for filename in files:
        if dog_detector(filename):
            count = count + 1
    return(count/total)

print(str(test_dog_detector(human_files_short)*100) + "% of human_files_short detected as dog")
print(str(test_dog_detector(dog_files_short)*100) + "% of dog_files_short detected as dog")
0.0% of human_files_short detected as dog
100.0% of dog_files_short detected as dog

We suggest VGG-16 as a potential network to detect dog images in your algorithm, but you are free to explore other pre-trained networks (such as Inception-v3, ResNet-50, etc). Please use the code cell below to test other pre-trained PyTorch models. If you decide to pursue this optional task, report performance on human_files_short and dog_files_short.

In [10]:
### (Optional) 
### TODO: Report the performance of another pre-trained network.
### Feel free to use as many code cells as needed.

Step 3: Create a CNN to Classify Dog Breeds (from Scratch)

Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 10%. In Step 4 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.

We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have trouble distinguishing between a Brittany and a Welsh Springer Spaniel.

Brittany Welsh Springer Spaniel

It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).

Curly-Coated Retriever American Water Spaniel

Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.

Yellow Labrador Chocolate Labrador Black Labrador

We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.

Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!

(IMPLEMENTATION) Specify Data Loaders for the Dog Dataset

Use the code cell below to write three separate data loaders for the training, validation, and test datasets of dog images (located at dogImages/train, dogImages/valid, and dogImages/test, respectively). You may find this documentation on custom datasets to be a useful resource. If you are interested in augmenting your training and/or validation data, check out the wide variety of transforms!

In [13]:
import os
from torchvision import datasets

### TODO: Write data loaders for training, validation, and test sets
## Specify appropriate transforms, and batch_sizes

num_workers = 0
batch_size = 128

normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])        

train_loader = torch.utils.data.DataLoader(
    datasets.ImageFolder('dogImages/train', transforms.Compose([
        transforms.Resize(256),
        transforms.CenterCrop(224),
        transforms.RandomHorizontalFlip(),
        transforms.RandomRotation(10),
        transforms.ToTensor(),
        normalize,
    ])), batch_size=batch_size, num_workers=num_workers, shuffle=True)

valid_loader = torch.utils.data.DataLoader(
    datasets.ImageFolder('dogImages/valid', transforms.Compose([
        transforms.Resize(256),
        transforms.CenterCrop(224),
        transforms.ToTensor(),
        normalize,
    ])), batch_size=batch_size, num_workers=num_workers, shuffle=True)

test_loader = torch.utils.data.DataLoader(
    datasets.ImageFolder('dogImages/test', transforms.Compose([
        transforms.Resize(256),
        transforms.CenterCrop(224),
        transforms.ToTensor(),
        normalize,
    ])), batch_size=batch_size, num_workers=num_workers, shuffle=True)

loaders_scratch = {'train': train_loader, 'test': test_loader, 'valid' : valid_loader}

Question 3: Describe your chosen procedure for preprocessing the data.

  • How does your code resize the images (by cropping, stretching, etc)? What size did you pick for the input tensor, and why?
  • Did you decide to augment the dataset? If so, how (through translations, flips, rotations, etc)? If not, why not?

Answer:

  • Random Resized Crop. I chose 224 because this is what the model expects
  • Data augmentation: Crop is randomized and I added a randon horizontal flip

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. Use the template in the code cell below.

In [19]:
import torch.nn as nn
import torch.nn.functional as F

# define the CNN architecture
class Net(nn.Module):
    ### TODO: choose an architecture, and complete the class
    def __init__(self):
        super(Net, self).__init__()
        ## Define layers of a CNN
        super(Net, self).__init__()
        ## Define layers of a CNN
        
        self.conv1 = nn.Conv2d(3, 16, 3, padding=1)
        self.norm1 = nn.BatchNorm2d(16)
        self.conv2 = nn.Conv2d(16, 32, 3, padding=1)
        self.norm2 = nn.BatchNorm2d(32)
        self.conv3 = nn.Conv2d(32, 64, 3, padding=1)
        self.norm3 = nn.BatchNorm2d(64)
        self.conv4 = nn.Conv2d(64, 64, 3, padding=1)
        self.norm4 = nn.BatchNorm2d(64)
        self.conv5 = nn.Conv2d(64, 64, 3, padding=1)
        self.norm5 = nn.BatchNorm2d(64)
        self.maxpool = nn.MaxPool2d(2,2)
        
        self.fc1 = nn.Linear(in_features=7*7*64, out_features=256)
        self.fc2 = nn.Linear(in_features=256, out_features=133)
        
        self.dropout = nn.Dropout(0.5)    
    def forward(self, x):
        ## Define forward behavior
        
        x = F.relu(self.norm1(self.conv1(x)))
        x = self.maxpool(x) # 224 => 112
        
        x = F.relu(self.norm2(self.conv2(x)))
        x = self.maxpool(x) # 112 => 56
        
        x = F.relu(self.norm3(self.conv3(x)))
        x = self.maxpool(x) # 56 => 28
       
        x = F.relu(self.norm4(self.conv4(x)))
        x = self.maxpool(x) # 28 => 14
        
        x = F.relu(self.norm5(self.conv5(x)))
        x = self.maxpool(x) # 14 => 7
        
        x = x.view(-1, 7 * 7 * 64)
            
        x = self.dropout(F.relu(self.fc1(x)))
        x = self.fc2(x)
        return x

#-#-# You do NOT have to modify the code below this line. #-#-#

# instantiate the CNN
model_scratch = Net()

# move tensors to GPU if CUDA is available
if use_cuda:
    model_scratch.cuda()

Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step.

Answer:

  • I created 4 layers of Relu(BatchNorm(Conv)) + max pool.
  • I increase the depth at each layer using convolution and I decrease the dimension using maxpooling.
  • Finally I added 2 fully connected layers + dropout to prevent overfitting.
  • It was inspired by VGG architecture.

(IMPLEMENTATION) Specify Loss Function and Optimizer

Use the next code cell to specify a loss function and optimizer. Save the chosen loss function as criterion_scratch, and the optimizer as optimizer_scratch below.

In [20]:
import torch.optim as optim

### TODO: select loss function
criterion_scratch = nn.CrossEntropyLoss()

### TODO: select optimizer
optimizer_scratch = optim.SGD(model_scratch.parameters(), lr=0.01)

(IMPLEMENTATION) Train and Validate the Model

Train and validate your model in the code cell below. Save the final model parameters at filepath 'model_scratch.pt'.

In [21]:
# the following import is required for training to be robust to truncated images
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

def train(n_epochs, loaders, model, optimizer, criterion, use_cuda, save_path):
    """returns trained model"""
    # initialize tracker for minimum validation loss
    valid_loss_min = np.Inf 
    
    for epoch in range(1, n_epochs+1):
        # initialize variables to monitor training and validation loss
        train_loss = 0.0
        valid_loss = 0.0
        
        ###################
        # train the model #
        ###################
        model.train()
        print('\ntraining')
        for batch_idx, (data, target) in enumerate(loaders['train']):
            if batch_idx % 10 == 0:
                print('.', end='')            
            # move to GPU
            if use_cuda:
                data, target = data.cuda(), target.cuda()
            ## find the loss and update the model parameters accordingly
            ## record the average training loss, using something like
            ## train_loss = train_loss + ((1 / (batch_idx + 1)) * (loss.data - train_loss))

            optimizer.zero_grad()
            output = model(data)
            loss = criterion(output, target)
            loss.backward()
            optimizer.step()
            
            train_loss = train_loss + ((1 / (batch_idx + 1)) * (loss.data - train_loss))             
            
        ######################    
        # validate the model #
        ######################
        model.eval()
        print('\nvalidation')
        for batch_idx, (data, target) in enumerate(loaders['valid']):
            if batch_idx % 10 == 0:
                print('.', end='')
            # move to GPU
            if use_cuda:
                data, target = data.cuda(), target.cuda()
            ## update the average validation loss

            with torch.no_grad():
                output = model(data)
                loss = criterion(output, target)
                valid_loss = valid_loss + ((1 / (batch_idx + 1)) * (loss.data - valid_loss))   
            
        # print training/validation statistics 
        print('\nEpoch: {} \tTraining Loss: {:.6f} \tValidation Loss: {:.6f}'.format(
            epoch, 
            train_loss,
            valid_loss
            ))
        
        ## TODO: save the model if validation loss has decreased
        if valid_loss <= valid_loss_min:
            print('Validation loss decreased ({:.6f} --> {:.6f}).  Saving model ...'.format(valid_loss_min, valid_loss))
            torch.save(model.state_dict(), save_path)
            valid_loss_min = valid_loss
            
    # return trained model
    return model
In [22]:
# train the model
model_scratch = train(100, loaders_scratch, model_scratch, optimizer_scratch, 
                      criterion_scratch, use_cuda, 'model_scratch.pt')

# load the model that got the best validation accuracy
model_scratch.load_state_dict(torch.load('model_scratch.pt'))
training
......
validation
.
Epoch: 1 	Training Loss: 4.905095 	Validation Loss: 4.869021
Validation loss decreased (inf --> 4.869021).  Saving model ...

training
......
validation
.
Epoch: 2 	Training Loss: 4.859314 	Validation Loss: 4.829437
Validation loss decreased (4.869021 --> 4.829437).  Saving model ...

training
......
validation
.
Epoch: 3 	Training Loss: 4.816763 	Validation Loss: 4.787287
Validation loss decreased (4.829437 --> 4.787287).  Saving model ...

training
......
validation
.
Epoch: 4 	Training Loss: 4.753685 	Validation Loss: 4.725108
Validation loss decreased (4.787287 --> 4.725108).  Saving model ...

training
......
validation
.
Epoch: 5 	Training Loss: 4.703694 	Validation Loss: 4.661097
Validation loss decreased (4.725108 --> 4.661097).  Saving model ...

training
......
validation
.
Epoch: 6 	Training Loss: 4.635853 	Validation Loss: 4.592792
Validation loss decreased (4.661097 --> 4.592792).  Saving model ...

training
......
validation
.
Epoch: 7 	Training Loss: 4.572369 	Validation Loss: 4.526126
Validation loss decreased (4.592792 --> 4.526126).  Saving model ...

training
......
validation
.
Epoch: 8 	Training Loss: 4.507665 	Validation Loss: 4.456288
Validation loss decreased (4.526126 --> 4.456288).  Saving model ...

training
......
validation
.
Epoch: 9 	Training Loss: 4.450255 	Validation Loss: 4.403354
Validation loss decreased (4.456288 --> 4.403354).  Saving model ...

training
......
validation
.
Epoch: 10 	Training Loss: 4.392022 	Validation Loss: 4.365706
Validation loss decreased (4.403354 --> 4.365706).  Saving model ...

training
......
validation
.
Epoch: 11 	Training Loss: 4.344954 	Validation Loss: 4.330766
Validation loss decreased (4.365706 --> 4.330766).  Saving model ...

training
......
validation
.
Epoch: 12 	Training Loss: 4.305188 	Validation Loss: 4.276318
Validation loss decreased (4.330766 --> 4.276318).  Saving model ...

training
......
validation
.
Epoch: 13 	Training Loss: 4.258913 	Validation Loss: 4.272979
Validation loss decreased (4.276318 --> 4.272979).  Saving model ...

training
......
validation
.
Epoch: 14 	Training Loss: 4.199089 	Validation Loss: 4.421492

training
......
validation
.
Epoch: 15 	Training Loss: 4.147466 	Validation Loss: 4.171160
Validation loss decreased (4.272979 --> 4.171160).  Saving model ...

training
......
validation
.
Epoch: 16 	Training Loss: 4.127861 	Validation Loss: 4.132199
Validation loss decreased (4.171160 --> 4.132199).  Saving model ...

training
......
validation
.
Epoch: 17 	Training Loss: 4.084201 	Validation Loss: 4.208014

training
......
validation
.
Epoch: 18 	Training Loss: 4.040882 	Validation Loss: 4.405077

training
......
validation
.
Epoch: 19 	Training Loss: 4.009370 	Validation Loss: 4.278907

training
......
validation
.
Epoch: 20 	Training Loss: 3.969815 	Validation Loss: 4.352092

training
......
validation
.
Epoch: 21 	Training Loss: 3.936912 	Validation Loss: 4.172953

training
......
validation
.
Epoch: 22 	Training Loss: 3.897939 	Validation Loss: 4.142834

training
......
validation
.
Epoch: 23 	Training Loss: 3.859968 	Validation Loss: 4.061660
Validation loss decreased (4.132199 --> 4.061660).  Saving model ...

training
......
validation
.
Epoch: 24 	Training Loss: 3.824694 	Validation Loss: 3.947563
Validation loss decreased (4.061660 --> 3.947563).  Saving model ...

training
......
validation
.
Epoch: 25 	Training Loss: 3.793600 	Validation Loss: 3.966495

training
......
validation
.
Epoch: 26 	Training Loss: 3.752395 	Validation Loss: 4.010270

training
......
validation
.
Epoch: 27 	Training Loss: 3.733542 	Validation Loss: 3.891510
Validation loss decreased (3.947563 --> 3.891510).  Saving model ...

training
......
validation
.
Epoch: 28 	Training Loss: 3.701876 	Validation Loss: 3.951382

training
......
validation
.
Epoch: 29 	Training Loss: 3.666593 	Validation Loss: 3.926934

training
......
validation
.
Epoch: 30 	Training Loss: 3.649792 	Validation Loss: 4.391729

training
......
validation
.
Epoch: 31 	Training Loss: 3.604307 	Validation Loss: 3.938869

training
......
validation
.
Epoch: 32 	Training Loss: 3.587246 	Validation Loss: 3.804670
Validation loss decreased (3.891510 --> 3.804670).  Saving model ...

training
......
validation
.
Epoch: 33 	Training Loss: 3.549890 	Validation Loss: 3.927794

training
......
validation
.
Epoch: 34 	Training Loss: 3.534566 	Validation Loss: 3.749712
Validation loss decreased (3.804670 --> 3.749712).  Saving model ...

training
......
validation
.
Epoch: 35 	Training Loss: 3.483562 	Validation Loss: 4.006981

training
......
validation
.
Epoch: 36 	Training Loss: 3.471230 	Validation Loss: 3.765831

training
......
validation
.
Epoch: 37 	Training Loss: 3.436392 	Validation Loss: 3.844686

training
......
validation
.
Epoch: 38 	Training Loss: 3.419639 	Validation Loss: 3.745631
Validation loss decreased (3.749712 --> 3.745631).  Saving model ...

training
......
validation
.
Epoch: 39 	Training Loss: 3.391981 	Validation Loss: 4.309092

training
......
validation
.
Epoch: 40 	Training Loss: 3.359370 	Validation Loss: 3.726540
Validation loss decreased (3.745631 --> 3.726540).  Saving model ...

training
......
validation
.
Epoch: 41 	Training Loss: 3.330766 	Validation Loss: 3.642903
Validation loss decreased (3.726540 --> 3.642903).  Saving model ...

training
......
validation
.
Epoch: 42 	Training Loss: 3.310943 	Validation Loss: 3.739939

training
......
validation
.
Epoch: 43 	Training Loss: 3.304453 	Validation Loss: 3.814581

training
......
validation
.
Epoch: 44 	Training Loss: 3.269016 	Validation Loss: 3.995353

training
......
validation
.
Epoch: 45 	Training Loss: 3.241758 	Validation Loss: 3.762702

training
......
validation
.
Epoch: 46 	Training Loss: 3.212077 	Validation Loss: 3.877168

training
......
validation
.
Epoch: 47 	Training Loss: 3.204113 	Validation Loss: 3.547238
Validation loss decreased (3.642903 --> 3.547238).  Saving model ...

training
......
validation
.
Epoch: 48 	Training Loss: 3.180858 	Validation Loss: 3.632342

training
......
validation
.
Epoch: 49 	Training Loss: 3.167224 	Validation Loss: 4.476603

training
......
validation
.
Epoch: 50 	Training Loss: 3.129692 	Validation Loss: 3.788965

training
......
validation
.
Epoch: 51 	Training Loss: 3.100790 	Validation Loss: 3.639293

training
......
validation
.
Epoch: 52 	Training Loss: 3.073003 	Validation Loss: 4.015180

training
......
validation
.
Epoch: 53 	Training Loss: 3.063008 	Validation Loss: 4.099971

training
......
validation
.
Epoch: 54 	Training Loss: 3.030502 	Validation Loss: 3.723237

training
......
validation
.
Epoch: 55 	Training Loss: 3.006017 	Validation Loss: 3.593438

training
......
validation
.
Epoch: 56 	Training Loss: 2.980469 	Validation Loss: 4.194561

training
......
validation
.
Epoch: 57 	Training Loss: 2.956903 	Validation Loss: 4.235017

training
......
validation
.
Epoch: 58 	Training Loss: 2.931721 	Validation Loss: 4.009234

training
......
validation
.
Epoch: 59 	Training Loss: 2.934429 	Validation Loss: 4.392602

training
......
validation
.
Epoch: 60 	Training Loss: 2.907235 	Validation Loss: 3.821007

training
......
validation
.
Epoch: 61 	Training Loss: 2.880637 	Validation Loss: 3.808053

training
......
validation
.
Epoch: 62 	Training Loss: 2.861947 	Validation Loss: 3.593778

training
......
validation
.
Epoch: 63 	Training Loss: 2.840393 	Validation Loss: 3.908182

training
......
validation
.
Epoch: 64 	Training Loss: 2.808216 	Validation Loss: 3.921686

training
......
validation
.
Epoch: 65 	Training Loss: 2.793643 	Validation Loss: 3.570217

training
......
validation
.
Epoch: 66 	Training Loss: 2.769732 	Validation Loss: 3.713683

training
......
validation
.
Epoch: 67 	Training Loss: 2.759849 	Validation Loss: 3.688850

training
......
validation
.
Epoch: 68 	Training Loss: 2.746974 	Validation Loss: 3.916181

training
......
validation
.
Epoch: 69 	Training Loss: 2.698058 	Validation Loss: 3.640626

training
......
validation
.
Epoch: 70 	Training Loss: 2.683845 	Validation Loss: 4.073072

training
......
validation
.
Epoch: 71 	Training Loss: 2.663502 	Validation Loss: 3.635195

training
......
validation
.
Epoch: 72 	Training Loss: 2.651813 	Validation Loss: 3.572980

training
......
validation
.
Epoch: 73 	Training Loss: 2.613867 	Validation Loss: 4.231155

training
......
validation
.
Epoch: 74 	Training Loss: 2.631937 	Validation Loss: 3.377319
Validation loss decreased (3.547238 --> 3.377319).  Saving model ...

training
......
validation
.
Epoch: 75 	Training Loss: 2.598651 	Validation Loss: 3.559217

training
......
validation
.
Epoch: 76 	Training Loss: 2.573088 	Validation Loss: 4.188218

training
......
validation
.
Epoch: 77 	Training Loss: 2.555469 	Validation Loss: 3.433543

training
......
validation
.
Epoch: 78 	Training Loss: 2.544823 	Validation Loss: 3.545171

training
......
validation
.
Epoch: 79 	Training Loss: 2.519765 	Validation Loss: 3.396890

training
......
validation
.
Epoch: 80 	Training Loss: 2.499808 	Validation Loss: 3.630838

training
......
validation
.
Epoch: 81 	Training Loss: 2.463888 	Validation Loss: 3.424301

training
......
validation
.
Epoch: 82 	Training Loss: 2.482479 	Validation Loss: 3.684169

training
......
validation
.
Epoch: 83 	Training Loss: 2.452804 	Validation Loss: 3.235852
Validation loss decreased (3.377319 --> 3.235852).  Saving model ...

training
......
validation
.
Epoch: 84 	Training Loss: 2.419806 	Validation Loss: 3.280951

training
......
validation
.
Epoch: 85 	Training Loss: 2.415849 	Validation Loss: 3.349214

training
......
validation
.
Epoch: 86 	Training Loss: 2.347914 	Validation Loss: 3.187692
Validation loss decreased (3.235852 --> 3.187692).  Saving model ...

training
......
validation
.
Epoch: 87 	Training Loss: 2.365545 	Validation Loss: 3.742942

training
......
validation
.
Epoch: 88 	Training Loss: 2.328645 	Validation Loss: 3.895718

training
......
validation
.
Epoch: 89 	Training Loss: 2.334997 	Validation Loss: 3.558115

training
......
validation
.
Epoch: 90 	Training Loss: 2.316249 	Validation Loss: 3.509640

training
......
validation
.
Epoch: 91 	Training Loss: 2.296947 	Validation Loss: 3.587544

training
......
validation
.
Epoch: 92 	Training Loss: 2.263070 	Validation Loss: 3.625595

training
......
validation
.
Epoch: 93 	Training Loss: 2.257669 	Validation Loss: 3.690243

training
......
validation
.
Epoch: 94 	Training Loss: 2.250704 	Validation Loss: 3.573315

training
......
validation
.
Epoch: 95 	Training Loss: 2.223307 	Validation Loss: 3.527866

training
......
validation
.
Epoch: 96 	Training Loss: 2.209362 	Validation Loss: 4.496181

training
......
validation
.
Epoch: 97 	Training Loss: 2.193487 	Validation Loss: 3.576167

training
......
validation
.
Epoch: 98 	Training Loss: 2.158954 	Validation Loss: 3.137414
Validation loss decreased (3.187692 --> 3.137414).  Saving model ...

training
......
validation
.
Epoch: 99 	Training Loss: 2.153384 	Validation Loss: 3.217638

training
......
validation
.
Epoch: 100 	Training Loss: 2.152055 	Validation Loss: 3.744803
Out[22]:
<All keys matched successfully>

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Use the code cell below to calculate and print the test loss and accuracy. Ensure that your test accuracy is greater than 10%.

In [23]:
def test(loaders, model, criterion, use_cuda):

    # monitor test loss and accuracy
    test_loss = 0.
    correct = 0.
    total = 0.

    model.eval()
    for batch_idx, (data, target) in enumerate(loaders['test']):
        # move to GPU
        if use_cuda:
            data, target = data.cuda(), target.cuda()
        # forward pass: compute predicted outputs by passing inputs to the model
        output = model(data)
        # calculate the loss
        loss = criterion(output, target)
        # update average test loss 
        test_loss = test_loss + ((1 / (batch_idx + 1)) * (loss.data - test_loss))
        # convert output probabilities to predicted class
        pred = output.data.max(1, keepdim=True)[1]
        # compare predictions to true label
        correct += np.sum(np.squeeze(pred.eq(target.data.view_as(pred))).cpu().numpy())
        total += data.size(0)
            
    print('Test Loss: {:.6f}\n'.format(test_loss))

    print('\nTest Accuracy: %2d%% (%2d/%2d)' % (
        100. * correct / total, correct, total))
In [24]:
# call test function    
test(loaders_scratch, model_scratch, criterion_scratch, use_cuda)
Test Loss: 3.147735


Test Accuracy: 24% (206/836)

Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)

You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.

(IMPLEMENTATION) Specify Data Loaders for the Dog Dataset

Use the code cell below to write three separate data loaders for the training, validation, and test datasets of dog images (located at dogImages/train, dogImages/valid, and dogImages/test, respectively).

If you like, you are welcome to use the same data loaders from the previous step, when you created a CNN from scratch.

In [25]:
## TODO: Specify data loaders
num_workers = 0
batch_size = 128

normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])        

data_transfer = datasets.ImageFolder('dogImages/train', transforms.Compose([
        transforms.Resize(256),
        transforms.CenterCrop(224),
        transforms.RandomHorizontalFlip(),
        transforms.RandomRotation(10),
        transforms.ToTensor(),
        normalize,
    ]))

data_transfer = { 'train' : data_transfer}

train_loader = torch.utils.data.DataLoader(data_transfer['train'], batch_size=batch_size, num_workers=num_workers, shuffle=True)

valid_loader = torch.utils.data.DataLoader(
    datasets.ImageFolder('dogImages/valid', transforms.Compose([
        transforms.Resize(256),
        transforms.CenterCrop(224),
        transforms.ToTensor(),
        normalize,
    ])), batch_size=batch_size, num_workers=num_workers, shuffle=True)

test_loader = torch.utils.data.DataLoader(
    datasets.ImageFolder('dogImages/test', transforms.Compose([
        transforms.Resize(256),
        transforms.CenterCrop(224),
        transforms.ToTensor(),
        normalize,
    ])), batch_size=batch_size, num_workers=num_workers, shuffle=True)

loaders_transfer = {'train': train_loader, 'test': test_loader, 'valid' : valid_loader}

(IMPLEMENTATION) Model Architecture

Use transfer learning to create a CNN to classify dog breed. Use the code cell below, and save your initialized model as the variable model_transfer.

In [26]:
import torchvision.models as models
import torch.nn as nn

## TODO: Specify model architecture 
model_transfer = models.vgg16(pretrained=True)
#model_transfer = models.resnet18(pretrained=True)

for param in model_transfer.features.parameters():
    param.requires_grad = False

n_inputs = model_transfer.classifier[6].in_features
last_layer = nn.Linear(n_inputs, 133)

model_transfer.classifier[6] = last_layer

if use_cuda:
    model_transfer = model_transfer.cuda()

Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.

Answer:

  • VGG16 already includes several classes for dogs. It should be therefore quite usefule for our task
  • I froze all the parameters in the features module
  • I change the classification module to match the number of class for our problem (133)

(IMPLEMENTATION) Specify Loss Function and Optimizer

Use the next code cell to specify a loss function and optimizer. Save the chosen loss function as criterion_transfer, and the optimizer as optimizer_transfer below.

In [38]:
criterion_transfer = nn.CrossEntropyLoss()
optimizer_transfer = optim.SGD(model_transfer.parameters(), lr=0.001)

(IMPLEMENTATION) Train and Validate the Model

Train and validate your model in the code cell below. Save the final model parameters at filepath 'model_transfer.pt'.

In [46]:
# train the model
model_transfer = train(15, loaders_transfer, model_transfer, optimizer_transfer, criterion_transfer, use_cuda, 'model_transfer.pt')

# load the model that got the best validation accuracy (uncomment the line below)
model_transfer.load_state_dict(torch.load('model_transfer.pt'))
training
......
validation
.
Epoch: 1 	Training Loss: 0.714371 	Validation Loss: 0.643404
Validation loss decreased (inf --> 0.643404).  Saving model ...

training
......
validation
.
Epoch: 2 	Training Loss: 0.699411 	Validation Loss: 0.631859
Validation loss decreased (0.643404 --> 0.631859).  Saving model ...

training
......
validation
.
Epoch: 3 	Training Loss: 0.681191 	Validation Loss: 0.639236

training
......
validation
.
Epoch: 4 	Training Loss: 0.677848 	Validation Loss: 0.644414

training
......
validation
.
Epoch: 5 	Training Loss: 0.661650 	Validation Loss: 0.637664

training
......
validation
.
Epoch: 6 	Training Loss: 0.652607 	Validation Loss: 0.629175
Validation loss decreased (0.631859 --> 0.629175).  Saving model ...

training
......
validation
.
Epoch: 7 	Training Loss: 0.640846 	Validation Loss: 0.634151

training
......
validation
.
Epoch: 8 	Training Loss: 0.643384 	Validation Loss: 0.642674

training
......
validation
.
Epoch: 9 	Training Loss: 0.650573 	Validation Loss: 0.635007

training
......
validation
.
Epoch: 10 	Training Loss: 0.620136 	Validation Loss: 0.627520
Validation loss decreased (0.629175 --> 0.627520).  Saving model ...

training
......
validation
.
Epoch: 11 	Training Loss: 0.624823 	Validation Loss: 0.611406
Validation loss decreased (0.627520 --> 0.611406).  Saving model ...

training
......
validation
.
Epoch: 12 	Training Loss: 0.634895 	Validation Loss: 0.615665

training
......
validation
.
Epoch: 13 	Training Loss: 0.617348 	Validation Loss: 0.597003
Validation loss decreased (0.611406 --> 0.597003).  Saving model ...

training
......
validation
.
Epoch: 14 	Training Loss: 0.649230 	Validation Loss: 0.610934

training
......
validation
.
Epoch: 15 	Training Loss: 0.596575 	Validation Loss: 0.618642
Out[46]:
<All keys matched successfully>

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Use the code cell below to calculate and print the test loss and accuracy. Ensure that your test accuracy is greater than 60%.

In [47]:
test(loaders_transfer, model_transfer, criterion_transfer, use_cuda)
Test Loss: 0.711727


Test Accuracy: 80% (677/836)

(IMPLEMENTATION) Predict Dog Breed with the Model

Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan hound, etc) that is predicted by your model.

In [48]:
### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.

# list of class names by index, i.e. a name can be accessed like class_names[0]
class_names = [item[4:].replace("_", " ") for item in data_transfer['train'].classes]

def predict_breed_transfer(img_path):
    # load the image and return the predicted breed
    transform = transforms.Compose([
        transforms.Resize(256),               
        transforms.CenterCrop(224),           
        transforms.ToTensor(),                
        transforms.Normalize(                 
            mean=[0.485, 0.456, 0.406],           
            std=[0.229, 0.224, 0.225]             
        )])
    
    image = Image.open(img_path)    
    image = transform(image)

    result = []     

    with torch.no_grad():
        model_transfer.eval()
    
        if use_cuda:
            image = image.cuda()
    
        batch = torch.unsqueeze(image, 0)

        output = model_transfer(batch) 
        output = F.softmax(output, dim=1) * 100.0
    
        topk, topclass = output.topk(3)

        topk = topk.cpu().numpy()[0]
        topclass = topclass.cpu().numpy()[0]

        for i in range(0,3):
            v = topk[i]
            if v > 10:
                result.append(class_names[topclass[i]] + " (" + str(round(v)) + "%)")
    return result

Step 5: Write your Algorithm

Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,

  • if a dog is detected in the image, return the predicted breed.
  • if a human is detected in the image, return the resembling dog breed.
  • if neither is detected in the image, provide output that indicates an error.

You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and dog_detector functions developed above. You are required to use your CNN from Step 4 to predict dog breed.

Some sample output for our algorithm is provided below, but feel free to design your own user experience!

Sample Human Output

(IMPLEMENTATION) Write your Algorithm

In [49]:
from IPython.display import display

### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.

def run_app(img_path):
    ## handle cases for a human face, dog, and neither
    if dog_detector(img_path):
        breeds = predict_breed_transfer(img_path)
        print("Hello Dog, you look like a ... ")
        for breed in breeds:
            print(breed)
    elif face_detector(img_path):
        breeds = predict_breed_transfer(img_path)
        if len(breeds) > 1:
            print("Hello Human, you look like a ... ")
            for breed in breeds:
                print(breed)
        else:
            print("Hello Human, you don't look like a dog.")
    else:
        print("I don't see anything")
    
    pil_im = Image.open(img_path, 'r')
    display(pil_im)

Step 6: Test Your Algorithm

In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?

(IMPLEMENTATION) Test Your Algorithm on Sample Images!

Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.

Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.

Answer: (Three possible points for improvement)

Output is as expected.

Possible points of improvements:

  • More augmentation during training
  • Crop around the dog/person because some pictures contain several dogs or a mix of dog and persons
  • Use another pre-trained network (Resnet? VGG19?)
In [50]:
## TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.

human_files = ['face1.jpg', 'face2.jpg', 'face3.jpg']
dog_files = ['dog1.jpg', 'dog2.jpg', 'dog3.jpg']

## suggested code, below
for file in np.hstack((human_files[:3], dog_files[:3])):
    run_app(file)
    print('------------------')
Hello Human, you don't look like a dog.
------------------
Hello Human, you don't look like a dog.
------------------
I don't see anything
------------------
Hello Dog, you look like a ... 
French bulldog (74.0%)
Bullmastiff (22.0%)
------------------
Hello Dog, you look like a ... 
Labrador retriever (96.0%)
------------------
Hello Dog, you look like a ... 
Collie (100.0%)
------------------